Nebius Group (NASDAQ: NBIS) is planning to raise prices on several of its on-demand GPU, CPU, and memory services starting October 1, according to a customer pricing notice reviewed by MarketCatalyst. The adjustments touch the company's most sought-after compute tiers — NVIDIA's H100, H200, B200, and B300 chips — and follow a strong second quarter shaped by rising AI Cloud demand.
For a company whose appeal to customers has largely rested on offering discounted capacity relative to the hyperscalers, a broad rate increase is worth noting. It may suggest that Nebius feels demand for its infrastructure is steady enough to comfortably absorb higher prices, without meaningfully affecting customer retention.
What's Changing
The notice lists new per-GPU-hour rates across four NVIDIA chip families, plus adjustments to underlying CPU and memory pricing:
| Hardware | New Rate | Previous Rate | Region(s) |
|---|---|---|---|
| NVIDIA H100 | $4.50 / GPU-hr | $3.85 / GPU-hr | eu-north1 |
| NVIDIA H200 | $5.40 / GPU-hr | $4.50 / GPU-hr | Listed European & U.S. regions |
| NVIDIA B200 | $8.50 / GPU-hr | $7.15 / GPU-hr | us-central1, me-west1 |
| NVIDIA B300 | $9.50 / GPU-hr | $7.85 / GPU-hr | uk-south1, eu-west2, us-north1 |
The increases range from roughly 15% for the H100 and H200 to about 21% for the B300, which sees the largest adjustment of the four. AMD EPYC Genoa CPU and memory rates are also moving slightly higher as part of the same notice, while Intel Ice Lake CPU pricing remains unchanged.
Why Nebius Is Raising Prices
Nebius describes the changes as part of scaling its global AI infrastructure and securing additional financing tied to deployed GPUs — language that points toward asset-backed financing structures, where lenders extend capital against contracted or installed hardware. In that context, pricing is likely about more than margin alone: higher effective revenue per GPU can also help support the economics that lenders and customers are underwriting against.
This also appears consistent with what Nebius has shared with investors more broadly. On its Q2 earnings call, management noted that the company has repeatedly sold out of available capacity as new supply comes online, and pointed to a capacity auction that cleared at a record price, roughly 15% above previous highs for Blackwell-generation chips. Taken together, these details suggest demand has been outpacing supply, and this notice may reflect Nebius adjusting its list prices accordingly.
The Q2 Backdrop
The price increases follow a second quarter in which Nebius's core cloud business scaled sharply:
Nebius has reaffirmed 2026 guidance calling for $3–$3.4 billion in group revenue, a roughly 40% adjusted EBITDA margin, and $20–$25 billion in capital expenditures as it continues building out data center capacity globally. Higher on-demand rates may offer the company an additional path toward those targets, without necessarily requiring capacity growth to move faster than planned.
What to Watch
The rate changes are set to take effect October 1 across the regions listed above. Customers with existing reserved or contracted capacity commitments are typically governed by separate contract terms rather than these on-demand list prices, though the notice offers a useful signal on the broader direction of AI compute costs as capacity remains tight across the industry.
The Takeaway
- This appears to be a pricing-power story rather than a distress signal. Every hardware tier seeing an increase is among Nebius's newest, most in-demand GPU capacity — the kind of asset that tends to see higher prices when demand outpaces supply.
- It's broadly consistent with the Q2 narrative. Sold-out capacity, strong auction clearing prices, and a 514% AI Cloud revenue surge all point in a similar direction: customer demand currently appears to be outpacing what Nebius can immediately supply.
- Financing and pricing seem increasingly connected. Tying new financing to deployed GPUs may make the economics of each GPU — including what it can be rented for — more directly relevant to Nebius's capital structure, and not just its income statement.
Frequently Asked Questions
When do Nebius's new GPU prices take effect?
According to a customer pricing notice, the new rates for H100, H200, B200, and B300 GPUs, along with AMD EPYC Genoa CPU and memory pricing, are set to take effect October 1.
How much are Nebius's GPU prices increasing?
H100 pricing in eu-north1 is set to rise to $4.50 per GPU-hour from $3.85. H200 pricing is set to rise to $5.40 per GPU-hour from $4.50 across listed European and U.S. regions. B200 pricing in us-central1 and me-west1 is set to rise to $8.50 from $7.15 per GPU-hour. B300 pricing in uk-south1, eu-west2, and us-north1 is set to rise to $9.50 from $7.85 per GPU-hour.
Why is Nebius raising GPU cloud prices?
Nebius appears to be scaling its global AI infrastructure footprint and has tied additional financing to deployed GPUs, while demand for its compute capacity has consistently outpaced supply — reflected in sold-out capacity and above-market clearing prices in recent capacity auctions.
How did Nebius perform financially before this price increase?
In Q2 2026, Nebius's AI Cloud revenue grew 514% year-over-year, and adjusted EBITDA moved to a $236.2 million profit from a $21 million loss a year earlier, on group revenue of $582.3 million, up 454% year-over-year.
This platform, including MarketCatalyst LLC, is not a registered investment advisor and doesn't manage client assets. Content here is for informational and educational purposes only — not investment advice, and not a stock-picking or trade-alert service. Trading stocks and options carries risk, including possible loss of principal. Consider your own goals, time horizon, and risk tolerance, and consult a qualified financial advisor before making any investment decisions.